Volume No: 6, Issue: 7, March 2013, e201303009, http://dx.doi.org/10.5936/csbj.201303009 CSBJ The albumin-binding domain as a scaffold for protein engineering Johan Nilvebrant a,*, Sophia Hober a Abstract: The albumin-binding domain is a small, three-helical protein domain found in various surface proteins expressed by gram-positive bacteria. Albumin binding is important in bacterial pathogenesis and several homologous domains have been identified. Such albumin-binding regions have been used for protein purification or immobilization. Moreover, improvement of the pharmacokinetics, through the non-covalent association to albumin, by fusing such domains to therapeutic proteins has been shown to be successful. Domains derived from streptococcal protein G and protein PAB from Finegoldia magna, which share a common origin and therefore represent an interesting evolutionary system, have been thoroughly studied structurally and functionally. Their albumin-binding sites have been mapped and these domains form the basis for a wide range of protein engineering approaches. By substitution-mutagenesis they have been engineered to achieve a broader specificity, an increased stability or an improved binding affinity, respectively. Furthermore, novel binding sites have been incorporated either by replacing the original albumin-binding surface, or by complementing it with a novel interaction interface. Combinatorial protein libraries, where several residues have been randomized simultaneously, have generated a large number of new variants with desired binding characteristics. The albumin- binding domain has also been utilized to explore the relationship between three-dimensional structure and amino acid sequence. Proteins with latent structural information built into their sequence, where a single amino acid substitution shifts the equilibrium in favor of a different fold with a new function, have been designed. Altogether, these examples illustrate the versatility of the albumin-binding domain as a scaffold for protein engineering. Introduction Many gram-positive bacteria express surface proteins with ability human serum albumin (HSA) adsorbed to bacteria could bind to and to bind serum proteins [1]. The surface proteins typically contain inactivate the antibacterial chemokine MIG/CXCL9 (monokine- tandemly repeated serum protein-binding domains with one or several induced by gamma-interferon/CXC ligand), which is released by specificities, which often include albumin binding [2,3]. The bacteria activated epithelium [10]. This albumin-dependent event protects can thereby camouflage themselves with bound host-proteins to evade from the antibacterial activity and promotes bacterial survival at the the immune system and potentially also scavenge protein-bound epithelium. Even though all functions of bacterial surface proteins are nutrients [4,5]. Albumin is the most abundant protein in plasma and not yet fully elucidated, they clearly provide the bacteria expressing expression of albumin-binding proteins has been shown to promote them with an evolutionary advantage. bacterial growth and virulence [5,6]. The bacterial species that express Streptococcal protein G (SPG), which binds to immunoglobulins albumin-binding domains are usually part of the normal human flora and albumins of several species, is expressed on the surface of certain and they are opportunistic pathogens. There are many different types streptococcal strains [11-13] and is one of the best-characterized of albumin-binding proteins with different size and function. For bacterial surface proteins. As indicated in figure 1, SPG from the example, more than 40 albumin-binding domains have been found in opportunistic streptococcal strain G148 has two functional regions one protein, forming a rod-like structure in a giant cell wall-associated containing three immunoglobulin-binding (C1-C3) and three fibronectin-binding molecule. This protein was found on the surface albumin-binding domains (ABD1-3), respectively [12,14]. The of Staphylococcus aureus and is called Ebh (ECM-binding protein immunoglobulin-binding domains share a common four-stranded homologue, Uniprot Q2FYJ6) [7,8]. These huge proteins, which beta-sheet fold with a single alpha helix packed onto the sheet (4ß+α) have also been found on streptococci (i.e. extracellular matrix-binding [15]. Of the three homologous albumin-binding domains, the C- protein (Embp), Uniprot Q8KQ73) [9], are in addition able to bind terminal ABD3 has been most extensively studied; it is referred to as fibronectin. They mediate adhesion and have been shown to be G148-ABD in the text and G148-ABD3 in figure 2A. Nuclear required for biofilm formation in vivo. An additional mechanism of magnetic resonance (NMR) spectroscopy has established that this 46 albumin binding was recently identified when it was shown that amino acid domain folds into a left-handed anti-parallel three-helix bundle (3α) [4,16], similar to the structure of the immunoglobulin- 1 binding domains of the well-studied staphylococcal protein A [17,18]. This structural element is found in several other proteins, aDivision of Protein Technology, School of Biotechnology, KTH Royal Institute which indicates that the 3α-fold is energetically and functionally of Technology, AlbaNova University Center, SE-106 91 Stockholm, Sweden favorable since it has been utilized broadly [19]. Interestingly, a structural evaluation of the repeating units in the giant albumin- * Corresponding author. binding protein Ebh showed that its domains, one of which is E-mail address: [email protected] (Johan Nilvebrant) responsible for albumin binding, are connected by a long helix that Engineered albumin-binding domains participates in two three helix bundles in two adjacent repeating units [8]. This helix is responsible for the global rod-like structure of the protein. Figure 2A. Sequence alignment of 16 homologous albumin-binding domains and two engineered variants. Conserved amino acids are shown in gray and differences are highlighted in color. G148-ABD3 and ALB8-GA (sequences 1 and 2) represent the best-studied domains. PSD-1 (sequence 17) is an engineered variant with broadened species specificity and ABDstable (sequence 18) is a variant that has been stabilized to alkaline treatment. The picture was generated in Geneious Pro version 5.5.7 created by Biomatters and is based on a similar picture by Johansson et al. [4]. Figure 1. Schematic representation of streptococcal protein G. Protein G consists of an N-terminal signal sequence (Ss), an albumin-binding region containing three albumin-binding domains and a C-terminal immunoglobulin-binding region. A spacer (S) separates the binding regions and a C-terminal sequence (W) anchors the protein to the cell wall. Various albumin-binding parts, BB, ABP and the smallest albumin-binding unit, the 46 amino acid albumin-binding domain (G148-ABD), are indicated. ABD folds into a stable three-helix bundle structure (the picture was generated from PDB-file 1GJT). Historically, the most widespread use of SPG has been as a biotechnological tool mainly used for affinity purification of immunoglobulins exploiting the broad species- and subclass specificity of its immunoglobulin-binding domains [20,21]. Albumin- binding regions spanning one or several albumin-binding domains, for example BB and ABP [22] that are indicated in figure 1, have been used for affinity purification or depletion of albumin [21]. Moreover, the use of an albumin-binding region as a fusion tag can facilitate Figure 2B. Structure of the complex formed by ALB8-GA and HSA. The affinity purification of a target protein, improve its solubility or be albumin-binding domains recognize a site located in domain II of HSA that used for directed immobilization [22-24]. does not overlap with the binding site for the neonatal Fc-receptor (FcRn), Several homologous albumin-binding domains have been which plays an important role in albumin homeostasis. The picture was identified in surface proteins from different bacterial species. The generated from PDB-file 1TF0. sequence diversity among these is illustrated by the 16 homologues included in Figure 2A. Alongside G148-ABD, the so-called protein Accumulated structural data on G148-ABD [4,16] and the GA- G-related albumin-binding (GA) module from protein PAB module [26-29] demonstrate that the domains have very similar (peptostreptococcal albumin-binding) of the anaerobic bacterium tertiary structures. ALB8-GA contains an additional residue in the Finegoldia magna (F. magna) has been thoroughly investigated both loop between the first and second helix (Figure 2A) and has a structurally and functionally [19,25,26]. Analysis of the gene somewhat shorter first helix compared to G148-ABD [4]. The encoding PAB suggested that its albumin-binding domain (ALB8-GA lengths and positions of the second and third helices are almost representing the best characterized variant, see Figure 2A) originates identical and this region also contains the most highly conserved from protein G and that it was introduced as a result of an sequence stretch among the homologues (Figure 2A), which implies interspecies module-shuffling event [25]. Available data on various that they all share a common overall fold. As would be expected, albumin-binding domains suggest a correlation between the species competitive binding studies have shown that G148-ABD and ALB8- specificity of the surface proteins and the host specificity of the GA have the same binding site on HSA [4]. A crystal